arXiv subjects

Christoph Weniger

Publications and source records attributed to Christoph Weniger.

At least 19 recordsLinked to original sources

arXiv2026

Fast Inference on Astronomical Time Series with Trans-Dimensional Flow Matching Posterior Estimation

The analysis of time series plays an important part in the study of (fast) transient events, including gamma-ray bursts, magnetar bursts, fast radio bursts, and solar flares. A common approach is to decompose the time series into pulses and study the pulse characteristics, such as location and amplitude, in order to constrain physical models of the source and its environment. However, estimating both the number and characteristics of these pulses presents a trans-dimensional inference problem that traditional sampling methods such as Markov Chain Monte Carlo (MCMC) and Nested Sampling struggle to solve efficiently. Simulation-based inference methods, often incorporating machine learning techniques, provide an alternative approach when traditional approaches are insufficient. Here, we introduce trans-dimensional Flow Matching Posterior Estimation (t-FMPE) implemented on a transformer architecture capable of efficient, amortized trans-dimensional inference on uniformly sampled univariate time series data. In this initial study, we apply the method to three test cases: simulated time series with known ground-truth parameters, observational data of Fast Radio Bursts and observations of X-ray bursts from magnetars. We show that t-FMPE achieves qualitative agreement with MCMC reference posteriors, successfully reproducing parameter correlations, as quantified through classifier two-sample tests. The trained network performs inference several orders of magnitude faster than MCMC and nested sampling, reaching sampling rates of 100 posterior samples per second for an 80-dimensional parameter space. The results demonstrate potential of t-FMPE for large-scale analysis of time series datasets when traditional sampling methods become infeasible, and also enable inferring unbiased posteriors in the presence of observational biases such as dead time.

astro-ph.IM↗

arXiv2025

PolySwyft: sequential simulation-based nested sampling

We present PolySwyft, a novel, non-amortised simulation-based inference framework that unites the strengths of nested sampling (NS) and neural ratio estimation (NRE) to tackle challenging posterior distributions when the likelihood is intractable but a forward simulator is available. By nesting rounds of NRE within the exploration of NS, and employing a principled KL-divergence criterion to adaptively terminate sampling, PolySwyft achieves faster convergence on complex, multimodal targets while rigorously preserving Bayesian validity. On a suite of toy problems with analytically known posteriors of a dim(theta,D)=(5,100) multivariate Gaussian and multivariate correlated Gaussian mixture model, we demonstrate that PolySwyft recovers all modes and credible regions with fewer simulator calls than swyft's TNRE. As a real-world application, we infer cosmological parameters dim(theta,D)=(6,111) from CMB power spectra using CosmoPower. PolySwyft is released as open-source software, offering a flexible toolkit for efficient, accurate inference across the astrophysical sciences and beyond.

astro-ph.IM↗

arXiv2025

How to embed any likelihood into SBI: Application to Planck + Stage IV galaxy surveys and Dynamical Dark Energy

Simulation-based inference (SBI) allows fast Bayesian inference for simulators encoding implicit likelihoods. However, some explicit likelihoods cannot be easily reformulated as simulators, hindering their integration into combined analyses within SBI frameworks. One key example in cosmology is given by the Planck CMB likelihoods. We present a simple method to construct an effective simulator for any explicit likelihood using samples from a previously converged Markov Chain Monte Carlo (MCMC) run. This effective simulator can subsequently be combined with any forward simulator. To illustrate this method, we combine the full Planck CMB likelihoods with a 3x2pt simulator (cosmic shear, galaxy clustering and their cross-correlation) for a Stage IV survey like Euclid, and test evolving dark energy parameterized by the $w_0w_a$ equation-of-state. Assuming the $w_0w_a$CDM cosmology hinted by DESI BAO DR2 + Planck 2018 + PantheonPlus SNIa datasets, we find that future 3x2pt data alone could detect evolving dark energy at $5σ$, while its combination with current CMB, BAO and SNIa datasets could raise the detection to almost $7σ$. Moreover, thanks to simulation reuse enabled by SBI, we show that our joint analysis is in excellent agreement with MCMC while requiring zero Boltzmann solver calls. This result opens up the possibility of performing massive global scans combining explicit and implicit likelihoods in a highly efficient way.

astro-ph.CO↗

arXiv2025

Dynamic SBI: Round-free Sequential Simulation-Based Inference with Adaptive Datasets

Simulation-based inference (SBI) is emerging as a new statistical paradigm for addressing complex scientific inference problems. By leveraging the representational power of deep neural networks, SBI can extract the most informative simulation features for the parameters of interest. Sequential SBI methods extend this approach by iteratively steering the simulation process towards the most relevant regions of parameter space. This is typically implemented through an algorithmic structure, in which simulation and network training alternate over multiple rounds. This strategy is particularly well suited for high-precision inference in high-dimensional settings, which are commonplace in physics applications with growing data volumes and increasing model fidelity. Here, we introduce dynamic SBI, which implements the core ideas of sequential methods in a round-free, asynchronous, and highly parallelisable manner. At its core is an adaptive dataset that is iteratively transformed during inference to resemble the target observation. Simulation and training proceed in parallel: trained networks are used both to filter out simulations incompatible with the data and to propose new, more promising ones. Compared to round-based sequential methods, this asynchronous structure can significantly reduce simulation costs and training overhead. We demonstrate that dynamic SBI achieves significant improvements in simulation and training efficiency while maintaining inference performance. We further validate our framework on two challenging astrophysical inference tasks: characterising the stochastic gravitational wave background and analysing strong gravitational lensing systems. Overall, this work presents a flexible and efficient new paradigm for sequential SBI.

astro-ph.IM↗

arXiv2025

Simulation-Based Inference of the sky-averaged 21-cm signal from CD-EoR with REACH

The redshifted 21-cm signal from the Cosmic Dawn and Epoch of Reionization carries invaluable information about the cosmology and astrophysics of the early Universe. Analyzing data from a sky-averaged 21-cm signal experiment requires navigating through an intricate parameter space addressing various factors such as foregrounds, beam uncertainties, ionospheric distortions, and receiver noise for the search of the 21-cm signal. The traditional likelihood-based sampling methods for modeling these effects could become computationally demanding for such complex models, which makes it infeasible to include physically motivated 21-cm signal models in the analysis. Moreover, the inference is driven by the assumed functional form of the likelihood. We demonstrate how Simulation-Based Inference through Truncated Marginal Neural Ratio Estimation (TMNRE) can naturally handle these issues at a reduced computational cost. We estimate the posterior distribution on our model parameters with TMNRE for simulated mock observations, incorporating beam-weighted foregrounds, physically motivated 21-cm signal, and radiometric noise. We find that maximizing information content by analyzing data from multiple time slices and antennas significantly improves the parameter constraints and enhances the exploration of the cosmological signal. We discuss the application of TMNRE for the current configuration of the REACH experiment and demonstrate its potential for exploring new avenues.

astro-ph.CO↗

arXiv2025

Sequential simulation-based inference for extreme mass ratio inspirals

Extreme mass-ratio inspirals pose a difficult challenge in terms of both search and parameter estimation for upcoming space-based gravitational-wave detectors such as LISA. Their signals are long and of complex morphology, meaning they carry a large amount of information about their source, but their waveforms are expensive to compute and they occupy a vast and multi-modal parameter space. We explore how sequential simulation-based inference methods, specifically truncated marginal neural ratio estimation, could offer solutions to some of the challenges surrounding extreme-mass-ratio inspiral data analysis. We show that this method can efficiently narrow down the volume of the complex 11-dimensional search parameter space by a factor of $10^6-10^7$ and provide 1-dimensional marginal proposal distributions for non-spinning extreme-mass-ratio inspirals. We discuss the current limitations of this approach and place it in the broader context of a global strategy for future space-based gravitational-wave data analysis.

gr-qc↗

arXiv2025

Leveraging Time-Dependent Instrumental Noise for LISA SGWB Analysis

Variations in the instrumental noise of the Laser Interferometer Space Antenna (LISA) over time are expected as a result of e.g. scheduled satellite operations or unscheduled glitches. We demonstrate that these fluctuations can be leveraged to improve the sensitivity to stochastic gravitational wave backgrounds (SGWBs) compared to the stationary noise scenario. This requires optimal use of data segments with downward noise fluctuations, and thus a data analysis pipeline capable of analysing and combining shorter time segments of mission data. We propose that simulation based inference is well suited for this challenge. In an approximate, but state-of-the-art, modeling setup, we show by comparison with Fisher Information Matrix estimates that the optimal information gain can be achieved in practice.

gr-qc↗

arXiv2025

A robust neural determination of the source-count distribution of the Fermi-LAT sky at high latitudes

Over the past 16 years, the Fermi Large Area Telescope (LAT) has significantly advanced our view of the GeV gamma-ray sky, yet several key questions remain - such as the composition of the isotropic gamma-ray background, the origin of the Fermi Bubbles or the potential presence of signatures from exotic physics like dark matter. Addressing these challenges requires sophisticated astrophysical modeling and robust statistical methods capable of handling high-dimensional parameter spaces. In this work, we analyze 14 years of high-latitude ($|b|\geq30^{\circ}$) Fermi-LAT data in the range from 1 to 10 GeV using simulation-based inference (SBI) via neural ratio estimation. This approach allows us to detect individual gamma-ray sources and derive a list of significant gamma-ray emitters containing more than 98\% of all sources listed in the Fermi-LAT Fourth Source Catalog (4FGL) with a flux $S>3\times10^{-10}\;\mathrm{cm}^{-2}\,\mathrm{s}^{-1}$ (about a factor of three larger than the flux above which 4FGL is nearly complete), without any non-4FGL source detected in that flux range. Additionally, we reconstruct the source-count distribution in both parametric and non-parametric forms, achieving large agreement with previous literature results as well as those sources detected by our SBI pipeline. We also quantitatively validate our gamma-ray emission simulator via an anomaly detection technique, demonstrating that the synthetic data closely reproduces the complexity of the real observations.

astro-ph.HE↗

arXiv2025

Strategic White Paper on AI Infrastructure for Particle, Nuclear, and Astroparticle Physics: Insights from JENA and EuCAIF

Artificial intelligence (AI) is transforming scientific research, with deep learning methods playing a central role in data analysis, simulations, and signal detection across particle, nuclear, and astroparticle physics. Within the JENA communities-ECFA, NuPECC, and APPEC-and as part of the EuCAIF initiative, AI integration is advancing steadily. However, broader adoption remains constrained by challenges such as limited computational resources, a lack of expertise, and difficulties in transitioning from research and development (R&D) to production. This white paper provides a strategic roadmap, informed by a community survey, to address these barriers. It outlines critical infrastructure requirements, prioritizes training initiatives, and proposes funding strategies to scale AI capabilities across fundamental physics over the next five years.

astro-ph.IM↗

arXiv2025

One never walks alone: the effect of the perturber population on subhalo measurements in strong gravitational lenses

Analyses of extended arcs in strong gravitational lensing images to date have constrained the properties of dark matter by measuring the parameters of one or two individual subhalos. However, since such analyses are reliant on likelihood-based methods like Markov-chain Monte Carlo or nested sampling, they require various compromises to the realism of lensing models for the sake of computational tractability, such as ignoring the numerous other subhalos and line-of-sight halos in the system, assuming a particular form for the source model and requiring the noise to have a known likelihood function. Here we show that a simulation-based inference method calledTruncated Marginal Neural Ratio Estimation (TMNRE) makes it possible to relax these requirements by training neural networks to directly compute marginal posteriors for subhalo parameters from lensing images. By performing a set of inference tasks on mock data, we verify the accuracy of TMNRE and show it can compute posteriors for subhalo parameters marginalized over populations of hundreds of substructures, as well as lens and source uncertainties. We also find the \gls*{mlp} Mixer network works far better for such tasks than the convolutional architectures explored in other lensing analyses. Furthermore, we show that since \gls*{tmnre} learns a posterior function it enables direct statistical checks that would be extremely expensive with likelihood-based methods. Our results show that TMNRE is well-suited for analyzing complex lensing data, and that the full subhalo and line-of-sight halo population must be included when measuring the properties of individual dark matter substructures with this technique.

astro-ph.CO↗

arXiv2025

Scalable inference with Autoregressive Neural Ratio Estimation

In recent years, there has been a remarkable development of simulation-based inference (SBI) algorithms, and they have now been applied across a wide range of astrophysical and cosmological analyses. There are a number of key advantages to these methods, centred around the ability to perform scalable statistical inference without an explicit likelihood. In this work, we propose two technical building blocks to a specific sequential SBI algorithm, truncated marginal neural ratio estimation (TMNRE). In particular, first we develop autoregressive ratio estimation with the aim to robustly estimate correlated high-dimensional posteriors. Secondly, we propose a slice-based nested sampling algorithm to efficiently draw both posterior samples and constrained prior samples from ratio estimators, the latter being instrumental for sequential inference. To validate our implementation, we carry out inference tasks on three concrete examples: a toy model of a multi-dimensional Gaussian, the analysis of a stellar stream mock observation, and finally, a proof-of-concept application to substructure searches in strong gravitational lensing. In addition, we publicly release the code for both the autoregressive ratio estimator and the slice sampler.

astro-ph.IM↗

arXiv2025

Tests for model misspecification in simulation-based inference: from local distortions to global model checks

Model misspecification analysis strategies, such as anomaly detection, model validation, and model comparison are a key component of scientific model development. Over the last few years, there has been a rapid rise in the use of simulation-based inference (SBI) techniques for Bayesian parameter estimation, applied to increasingly complex forward models. To move towards fully simulation-based analysis pipelines, however, there is an urgent need for a comprehensive simulation-based framework for model misspecification analysis. In this work, we provide a solid and flexible foundation for a wide range of model discrepancy analysis tasks, using distortion-driven model misspecification tests. From a theoretical perspective, we introduce the statistical framework built around performing many hypothesis tests for distortions of the simulation model. We also make explicit analytic connections to classical techniques: anomaly detection, model validation, and goodness-of-fit residual analysis. Furthermore, we introduce an efficient self-calibrating training algorithm that is useful for practitioners. We demonstrate the performance of the framework in multiple scenarios, making the connection to classical results where they are valid. Finally, we show how to conduct such a distortion-driven model misspecification test for real gravitational wave data, specifically on the event GW150914.

astro-ph.IM↗

arXiv2025

JENA Computing Initiative WP2 Report: Software and Heterogeneous Architectures

The scientific communities of nuclear, particle, and astroparticle physics are continuing to advance and are facing unprecedented software challenges due to growing data volumes, complex computing needs, and environmental considerations. As new experiments emerge, software and computing needs must be recognised and integrated early in design phases. This document synthesises insights from ECFA, NuPECC and APPEC, representing particle physics, nuclear physics, and astroparticle physics, and presents collaborative strategies for improving software, computing frameworks, infrastructure, and career development within these fields.

physics.comp-ph↗

arXiv2025

Fast Sampling of Cosmological Initial Conditions with Gaussian Neural Posterior Estimation

Knowledge of the primordial matter density field from which the large-scale structure of the Universe emerged over cosmic time is of fundamental importance for cosmology. However, reconstructing these cosmological initial conditions from late-time observations is a notoriously difficult task, which requires advanced cosmological simulators and sophisticated statistical methods to explore a multi-million-dimensional parameter space. We show how simulation-based inference (SBI) can be used to tackle this problem and to obtain data-constrained realisations of the primordial dark matter density field in a simulation-efficient way with general non-differentiable simulators. Our method is applicable to full high-resolution dark matter $N$-body simulations and is based on modelling the posterior distribution of the constrained initial conditions to be Gaussian with a diagonal covariance matrix in Fourier space. As a result, we can generate thousands of posterior samples within seconds on a single GPU, orders of magnitude faster than existing methods, paving the way for sequential SBI for cosmological fields. Furthermore, we perform an analytical fit of the estimated dependence of the covariance on the wavenumber, effectively transforming any point-estimator of initial conditions into a fast sampler. We test the validity of our obtained samples by comparing them to the true values with summary statistics and performing a Bayesian consistency test.

astro-ph.CO↗

arXiv2025

Large Physics Models: Towards a collaborative approach with Large Language Models and Foundation Models

This paper explores ideas and provides a potential roadmap for the development and evaluation of physics-specific large-scale AI models, which we call Large Physics Models (LPMs). These models, based on foundation models such as Large Language Models (LLMs) - trained on broad data - are tailored to address the demands of physics research. LPMs can function independently or as part of an integrated framework. This framework can incorporate specialized tools, including symbolic reasoning modules for mathematical manipulations, frameworks to analyse specific experimental and simulated data, and mechanisms for synthesizing theories and scientific literature. We begin by examining whether the physics community should actively develop and refine dedicated models, rather than relying solely on commercial LLMs. We then outline how LPMs can be realized through interdisciplinary collaboration among experts in physics, computer science, and philosophy of science. To integrate these models effectively, we identify three key pillars: Development, Evaluation, and Philosophical Reflection. Development focuses on constructing models capable of processing physics texts, mathematical formulations, and diverse physical data. Evaluation assesses accuracy and reliability by testing and benchmarking. Finally, Philosophical Reflection encompasses the analysis of broader implications of LLMs in physics, including their potential to generate new scientific understanding and what novel collaboration dynamics might arise in research. Inspired by the organizational structure of experimental collaborations in particle physics, we propose a similarly interdisciplinary and collaborative approach to building and refining Large Physics Models. This roadmap provides specific objectives, defines pathways to achieve them, and identifies challenges that must be addressed to realise physics-specific large scale AI models.

physics.data-an↗

arXiv2024

Fast likelihood-free inference in the LSS Stage IV era

Forthcoming large-scale structure (LSS) Stage IV surveys will provide us with unprecedented data to probe the nature of dark matter and dark energy. However, analysing these data with conventional Markov Chain Monte Carlo (MCMC) methods will be challenging, due to the increase in the number of nuisance parameters and the presence of intractable likelihoods. In light of this, we present the first application of Marginal Neural Ratio Estimation (MNRE) (a recent approach in simulation-based inference) to LSS photometric probes: weak lensing, galaxy clustering and the cross-correlation power spectra. In order to analyse the hundreds of spectra simultaneously, we find that a pre-compression of data using principal component analysis, as well as parameter-specific data summaries lead to highly accurate results. Using expected Stage IV experimental noise, we are able to recover the posterior distribution for the cosmological parameters with a speedup factor of $\sim 10-60$ compared to classical MCMC methods. To illustrate that the performance of MNRE is not impeded when posteriors are highly non-Gaussian, we test a scenario of two-body decaying dark matter, finding that Stage IV surveys can improve current bounds on the model by up to one order of magnitude. This result supports that MNRE is a powerful framework to constrain the standard cosmological model and its extensions with next-generation LSS surveys.

astro-ph.CO↗

arXiv2024

Radio sensitivity to a new population of millisecond pulsars in the Sagittarius Dwarf Spheroidal Galaxy

Observations with the Fermi Gamma-Ray Space Telescope reveal an excess of extended gamma-ray emission likely caused by an undiscovered population of millisecond pulsars (MSPs) in the core of the Sagittarius dwarf spheroidal galaxy (Sgr dSph). However, additional evidence, such as multi-wavelength searches, is necessary to confirm this theory. A significant discovery could be made if radio pulsations from individual MSPs in the Sgr dSph are detected. In this study, we investigate the possibility of detecting MSPs in the Sgr dSph with present and upcoming radio surveys using a phenomenological model based on the observed luminosity function of MSPs in the Milky Way's globular clusters. Our findings suggest that the Square Kilometer Array (SKA) is the most sensitive instrument for detecting these objects. We demonstrate that to observe one MSP with MeerKAT, we would need to perform a pointing observation of the core region of the Sgr dSph for about two hours. In this same observation time, SKA can identify $9^{+5}_{-3}$ MSPs in the entire system. Based on the distance of the Sgr dSph galaxy and our dispersion measure distance estimate, we find it possible to differentiate between MSPs belonging to the Sgr dSph and those of the Galactic disk and bulge. Furthermore, the MSPs hypothesis for the Sgr dSph gamma-ray excess could be confirmed at the 99.7\% confidence level by detecting at least six MSPs in a two-hour SKA observation of the Sgr dSph.

astro-ph.HE↗

arXiv2024

Mean-Field Simulation-Based Inference for Cosmological Initial Conditions

Reconstructing cosmological initial conditions (ICs) from late-time observations is a difficult task, which relies on the use of computationally expensive simulators alongside sophisticated statistical methods to navigate multi-million dimensional parameter spaces. We present a simple method for Bayesian field reconstruction based on modeling the posterior distribution of the initial matter density field to be diagonal Gaussian in Fourier space, with its covariance and the mean estimator being the trainable parts of the algorithm. Training and sampling are extremely fast (training: $\sim 1 \, \mathrm{h}$ on a GPU, sampling: $\lesssim 3 \, \mathrm{s}$ for 1000 samples at resolution $128^3$), and our method supports industry-standard (non-differentiable) $N$-body simulators. We verify the fidelity of the obtained IC samples in terms of summary statistics.

astro-ph.CO↗